Architecture and enhanced-algorithms to manage servers-processes into network: a management system.

Area of Science:

• Computer Science
• Operations Research
• Algorithm Design

Background:

• Minimizing makespan on identical parallel processors is a critical NP-hard scheduling problem.
• Existing methods struggle with efficiency and solution optimality for large-scale server networks.

Purpose of the Study:

• To develop novel algorithms for minimizing makespan in multi-server environments with identical parallel processors.
• To propose a new network architecture incorporating server management for enhanced scheduling.

Main Methods:

• Developed two novel heuristics based on problem decomposition into sub-problems.
• Employed a tree-based approach, saving intermediate results for subsequent decomposition levels.
• Introduced a new network architecture with server management.

Main Results:

• The proposed heuristics demonstrated significantly faster running times compared to literature rivals.
• The Most Loaded and Least Loaded (MLS) heuristic achieved the best solution in 87.9% of instances.
• The best heuristic found the optimal solution in 87.4% of cases, averaging 0.002s versus 1.307s for the best competitor.

Conclusions:

• The novel algorithms offer a highly efficient and effective solution for the NP-hard makespan minimization problem.
• The proposed methods are particularly beneficial for network scheduling scenarios with multiple servers.
• Experimental validation confirms superior performance in terms of speed and solution accuracy.